AskAI.free

Use ChatGPT, Claude and DeepSeek via AskAI

Use the authenticated AskAI.free REST API at https://askai.free. It provides a single interface to the ChatGPT, Claude, and DeepSeek models enabled for the user's AskAI plan.

Before the first request

Require ASKAI_API_KEY. If it is missing, tell the user how to create one; do not ask them to paste the secret into chat.

1. Sign in at https://askai.free/account.
2. Open the API Access card and create a key.
3. Copy it when shown and set it locally as ASKAI_API_KEY.

API keys are available only to verified, paying Pro and Max accounts after the first successful subscription payment. A free trial does not include API access. API calls spend the same monthly token allowance as the AskAI website; there is no separate API balance.

Treat the key as a secret. Never print it, include it in a URL, commit it, or send it to any host other than https://askai.free. Use the Authorization: Bearer $ASKAI_API_KEY header.

Workflow

1. If the user did not give an exact model id, call GET /v1/models. Choose an entry with available: true; prefer chatgpt-5-6-sol when it is present, otherwise use the first suitable available model.
2. Build a messages array from the user's request. Preserve their text and intent. Add a system message only when it materially helps or the user requested one.
3. Call POST /v1/chat/completions with stream: false. Use max_tokens: 1000 by default, increasing it only when the requested output needs more detail.
4. Return choices[0].message.content. Do not expose the authorization header or unnecessary raw response metadata.
5. Call GET /v1/usage only when the user asks about their plan or remaining allowance, or when diagnosing an insufficient-token response.

Endpoints

List models

bash

curl --fail-with-body --silent --show-error \
  https://askai.free/v1/models \
  -H "Authorization: Bearer $ASKAI_API_KEY"

The response contains data[] entries with id, name, provider, token_multiplier, and available. Do not select a model whose available value is false.

Create a chat completion

bash

curl --fail-with-body --silent --show-error \
  https://askai.free/v1/chat/completions \
  -H "Authorization: Bearer $ASKAI_API_KEY" \
  -H "Content-Type: application/json" \
  --data-binary @request.json

Write request.json as valid JSON rather than interpolating unescaped user text into a shell command:

json

{
  "model": "chatgpt-5-6-sol",
  "messages": [
    {"role": "user", "content": "The user's request"}
  ],
  "max_tokens": 1000,
  "stream": false
}

Delete temporary request files after use when the environment does not manage them automatically. The response follows the OpenAI chat-completions shape. Usage includes askai_tokens_charged and askai_tokens_remaining.

Check usage

bash

curl --fail-with-body --silent --show-error \
  https://askai.free/v1/usage \
  -H "Authorization: Bearer $ASKAI_API_KEY"

This returns the plan, remaining tokens, monthly allowance, and rate limit. Model listing and usage checks do not consume model tokens.

Errors and limits

• 401 missing_api_key or invalid_api_key: explain that the local key is missing, wrong, or revoked. Do not repeat it.
• 403 trial: API access begins after the first successful subscription payment.
• 403 not_pro: the account needs an active Pro or Max plan.
• 404 model_not_found: refresh /v1/models and choose a current model id.
• 413 context_length_exceeded: shorten or split the input.
• 429 insufficient_tokens: check /v1/usage; do not retry repeatedly.
• 429 ratelimit: wait before one bounded retry.
• 503 model_unavailable: choose another available model or report the temporary outage.

Streaming is not supported. Do not use this API for image generation, media-library browsing, subscription changes, or destructive account operations. Direct users to https://askai.free/api for the maintained API documentation.